How to Assign Tasks to Team Members Without Losing Track

To assign tasks to team members without losing track, give every task exactly one named owner, pick that owner by skill and by how much they already carry, write down what done looks like before they start, and agree when you will next look at it together. Then record all four where the whole team can see them, not in a private chat. Most lost tasks were never really assigned: they were mentioned in a meeting, two people each thought the other had them, or the owner was never told what finished meant. The same steps work when the owner is an AI assistant, with tighter limits.

This guide is about the day-to-day act of putting a name on a task. Handing over a whole area of work, and how much authority goes with it, is covered in how to delegate tasks. Mapping who does, approves, advises and hears about each piece of a project is the job of a RACI matrix.

One owner per task

A task with two owners has none. Each assumes the other is moving it, and it sits. A task with a team as its owner (“design will look at it”) is worse, because nobody in particular has to answer when someone asks where it is. Atlassian’s roles and responsibilities play↗ (opens in a new tab) includes a step to “Identify unassigned responsibilities” for exactly this reason: the work that falls through is usually work that belonged to everyone.

• The owner is the person who moves the task and answers “where is it?” Other people can help; the owner is the one name you would ask.
• If a task genuinely needs two people, split it into two tasks, each with one owner, and link them.
• If nobody can own it this week, say so and leave it unassigned in the backlog. An honest “nobody” beats a name on a task that person never agreed to.
• Assign the person, not the role. “Whoever is on support” is fine as a rota, but each task still needs today’s name on it.

Match the task to skill and capacity

The best person for a task is not always the one who would do it fastest. Ask three questions before you put a name on it:

1. Who can do this well enough? Fastest is only one answer. A task a little beyond someone is how they grow, as long as you allow the extra time and say that you are doing it on purpose.
2. Who has room? Count what each person already has in progress, not what they have in the backlog. Someone with four tasks in progress does not have room for a fifth, however quick it looks.
3. Who already has the context? The person who fixed the last payment bug will find the next one in half the time. Use that for urgent work; spread it around for everything else, so the knowledge does not live in one head.

Ask before you assign, when you can. “Can you take the export bug this week, or is your plate full?” gets you a real answer about capacity, and an owner who agreed rather than one who was told.

Say what done looks like

The first item in Gallup’s Q12 engagement survey↗ (opens in a new tab) is “I know what is expected of me at work.” At the level of a single task, that means two to five statements someone else could check. “Fix the export” is a topic. “Exporting a board with 500 tasks finishes in under ten seconds, and the file opens in Excel” is done.

Write the done-when with the owner, not for them. They will spot the case you missed, and a definition they helped write is one they will hold themselves to. For work that recurs, a team-wide definition of done saves writing the same checks every time.

A hand-off message template

An assignment is short. It is a message, not a brief, and it should fit on a phone screen. Post it where the task lives, so the next person who opens the task sees it too.

Task:       [ref and one-line title]
Owner:      [one name]
Why you:    [skill, context, or "a stretch, and here is why"]
Why now:    [what it unblocks, or who is waiting]
Done when:  [2-5 checkable statements]
Priority:   [number, and what it displaces if anything]
Check-in:   [when, and how: a comment, a message, a call]
Ask:        [who to go to when stuck]

Task:       BUG-212 Board export times out on large boards
Owner:      Priya
Why you:    You wrote the export; nobody else knows the query.
Why now:    Two customers hit it this week.
Done when:  - 500-task export finishes in under 10 seconds
            - The file opens in Excel and Numbers
            - How it was tested is written on the task
Priority:   2. It goes ahead of the settings redesign.
Check-in:   A comment on the task Thursday with where it stands.
Ask:        Sam, if the fix needs a database change.

Checking in without micromanaging

Checking in is part of assigning. The Office of Personnel Management’s guidance on monitoring performance↗ (opens in a new tab) describes the purpose well: ongoing monitoring “provides the opportunity to check how employees are doing and to identify and resolve any problems early.” Early is the point. Micromanaging is checking the method; a good check-in asks about the result and the obstacles.

• Check in when you said you would, and not otherwise. Unplanned check-ins read as distrust.
• Ask “What is in your way?” rather than “How far along are you?” The first gets you something you can fix.
• Read the task before you ask. If the owner has already written where it stands, a question you could have answered yourself costs them time.
• Make it cheap to say “I am stuck.” The owner who raises a problem on day two has done you a favor, compared with the one who raises it on the due date.

Assigning tasks to AI assistants

An AI assistant can own the doing of a task, but a person still answers for it. Give it a task the same way, with more written down, because it cannot ask the hallway question. Anthropic’s Claude Code best practices↗ (opens in a new tab) put it bluntly: “Claude can infer intent, but it can’t read your mind.” How to write a single task an assistant can finish is covered in giving an AI agent a task it can finish; the assigning habits are these:

• Assign it work with a check it can run: a test, a build, a page to compare. Without one, “looks done” is the only signal it has.
• Name the person accountable for the assistant’s task, in the task. That person reviews the evidence before the task counts as done.
• Keep an assistant’s tasks small and separate from people’s. If a person and an assistant both work on one task, split it.
• Write the limits down: what it must not change, and when to stop and ask. How work passes between people and assistants mid-task is covered in handing work between AI agents and people.

Reassigning when plans change

Priorities move, people get sick, and a task sometimes turns out to need a different skill. Reassigning is normal. Doing it silently is what loses work.

1. Tell the current owner first, and say why. “The API change needs Sam” is information; a task that vanishes from someone’s list is a message you did not mean to send.
2. Ask the current owner for a short handover on the task: what is done, what was tried, and what is left.
3. Confirm the new owner has room, the same way you would for a new task. Reassigning onto a full plate just moves the problem.
4. Change the name where the task lives, and add a line saying who had it before. The next person who asks will need both.
5. If the task no longer matters, close it with a reason rather than reassigning it.

Showing who owns what on a fenbs board

fenbs has no assignee field you can set, so ownership is shown with things the board already has. The habits below are honest about that and work well for a small team:

• Put the owner’s name in the first line of the task’s note, or at the start of the title, such as “Priya: board export times out.” Search covers both.
• Use a category per person if you want a filtered view of each person’s work. Categories are optional labels you create by typing a name, and the board filters by them. Keep projects for products or clients, because each project’s rules and context notes are given to AI assistants working on that project.
• History records who filed, changed and moved each task, so you can see who has actually been working on it, whatever the note says.
• Follow this task sends you an email when the task moves or someone comments, which is a check-in that costs the owner nothing.

For an AI assistant there is a direct form of assignment. On a task with a plan, a person with All Access presses “Let AI do this”, optionally with a limits line, and any connected assistant can take it with fenbs_next_approved_task. The task is held for that assistant while it works, so two assistants never take the same one, and History records its changes under its own name, such as “Claude via Priya”.